Social Media Analytics of Men's & Women's Clothing Store Reviews
1. Social Media Analytics & Web Mining
(MANGT 670)
Matthew Heinrich
Sarif Patwary
Brandon Payne
2. Business Problem & Approach
We wanted to explore clothing store reviews from Yelp:
Men’s Clothing stores & reviews
Women’s Clothing stores & reviews
3. Business Problem & Approach
From the review data, we wanted to explore the following:
• Sentiment Analysis
• Descriptive Text and Networks
• Topic Modeling
4. Data Source & Extraction
We used the Yelp dataset from here:
https://www.yelp.com/dataset_challenge
Our dataset contained:
-Businesses & Reviews
-Data from 10 cities, across 4 countries
-JSON Format
5. Sentiment Analysis
Wanted to find:
• Sentiment on text review using Bing Liu
• Sentiment on text review using Pattern
• Sentiment from Star Review
• Comparison of sentiment between men and women
6. Sentiment Analysis - Men’s Clothing Reviews
• About 80% (Bing Liu) to 86% (Pattern) men writes positive review whereas 70% gives 4 to 5 star
for a particular review
• About 12% (Pattern) to 13% (Bing Liu) men writes negative review whereas 24% gives 1 to 2
star for a particular review
• About 1% (Pattern) to 2% (Bing Liu) men writes neutral review whereas 6% gives exactly 3 star
review
7. Sentiment Analysis - Women’s Clothing Reviews
• About 80% (Bing Liu) to 86% (Pattern) women writes positive review whereas 64% gives 4 star
or 5 star for a particular review
• About 13% ( both Bing Liu & Pattern) women writes negative review whereas 23% gives 1 to 2
star for a particular review
• About 1% (Pattern) to 7% (Bing Liu) women writes neutral review whereas 12% gives exactly 3
star review
8. Sentiment Analysis - Conclusion
Both men and women show similar sentiment in text review.
Both men and women are highly positive in text review.
Star review reveals more negative sentiment than the text review for both
Women reviewers give slightly lower rating in star review than men.
9. Descriptive Text - Men’s Clothing Reviews
Collected 1,713 Reviews
Popular Review Locations:
Count’s Kustom: 83 Reviews
Artful Tailoring: 42 Reviews
Undefeated: 42 Reviews
Most Popular Words:
Store: 1,046 Words
Suit: 807 Words
Great: 799 Words
11. Descriptive Text - Women’s Clothing Reviews
Collected 7,650 Reviews
Popular Review Locations:
Anthropologie: 133 Reviews
Lululemon Athletica: 101 Reviews
Kate Spade New York: 54 Reviews
Most Popular Words:
Store: 6695 Words
Love: 2622 Words
Dress: 1374 Words
15. Topic Modeling - Men’s Clothing Stores
People who shop at Men's Clothing Stores are concerned with:
Selection of suits, shirts, shoes, boots, etc.
Customer Service (Good/Bad)
Weddings and Events
Customization of clothing (tailoring)
Quality of the clothing
Price - sales vs. online stores
16. Topic Modeling - Women’s Clothing Stores
People who shop at Women's Clothing Stores are concerned with:
The Brands that the stores offer.
Customer Service (Good/Bad) and Returns
Accessories like wallets, bags, and jewelry
Customization of clothing (tailoring) and how clothes fit
Quality of the clothing
Price
Store location
17. Overall Conclusions
What are overall sentiments toward Men's and Women's clothing stores?
Overall, Men's and Women's Clothing Store reviews were positive.
Are Men's and Women's sentiments different? If so, how?
Not really. The percentage of positive reviews for Men's stores vs. Women's
stores is nearly identical.
18. Overall Conclusions
How do "Star Reviews" compare to the sentiment from "Text Reviews"?
Star Reviews are typically more negative (less positive) than actual text
reviews.
What are popular words/topics?
Brands, types of clothes, customer service... (more in Topic Modeling)
Would focus on these areas if we were a particular Store/Brand